The rapid development of artificial intelligence is helping to reshape sectors such as education, healthcare, the environment, and the workplace, and contributing to economic and social transformation.
However, its ongoing development and deployment comes saddled with environmental costs.
The UN Regional Information Centre for Western Europe recently published a report, Artificial intelligence: How much energy does AI use?, which explored some of the ways in which AI is impacting the environment. Let’s look at these in more detail.
1. Energy consumption
Chief among the environmental concerns associated with AI is energy consumption. AI models, especially large language models, or LLMs, require significant computing power, in turn leading to a surge in electricity use – in Ireland, data centres accounted for 17% of the country’s electricity usage in 2022, with this demand expected to double by 2026.
According to the International Energy Agency (IEA), a single ChatGPT request requires ten times more electricity than a Google search.
2. Water consumption
While water is not necessarily the first thing you might think of when talking about AI, the data centres that power this next-generation technology need to keep cool, hence the need for water as an essential resource.
Global AI-related water use is projected to reach between 4.2 billion and 6.6 billion cubic meters by 2027, which would surpass Denmark’s total annual water withdrawal of between 4 billion and 6 billion cubic metres, according to the UN.
3. Raw materials
While AI is typically located in the ‘cloud’, it is nonetheless reliant on physical resources and raw materials. The minerals and rare earth elements required to develop servers and data centres are often extracted through mining processes that can cause deforestation, pollution, and land degradation.
According to the UN, the manufacturing of a typical, two-kilogram computer can require as much as 800 kilograms of raw materials.
4. Emissions
While machine learning and AI accounted for less than 0.1% of global GHG emissions in 2021, the rapid expansion of the technology in the years since has led to concerns over the emissions potential of this nascent industry.
As AI use continues to grow, the use of low-carbon energy sources to power it will become even more of an imperative, in order to reduce or maintain GHG emissions.
5. Electronic waste
AI also has the potential to contribute to electronic waste. As technology advances, older hardware becomes obsolete and is discarded, with this e-waste often containing hazardous substances such as lead and mercury.
Without proper recycling and disposal systems, this waste can pollute ecosystems and pose risks to human health.
At the same time, of course, AI can be a useful ally in supporting environmental solutions, enabling both governments and private enterprises to better monitor pollution, manage energy systems more efficiently, and forecast climate-related risks.
As the UN Environment Programme’s Climate Technology Progress Report 2024 noted, AI’s role in mapping renewable energy potential, optimising efficiency, and facilitating interconnectivity with other sectors, such as water and agriculture, is invaluable.
‘However, AI cannot fully replace the physical infrastructure and governance systems essential for the energy transition,’ it said. ‘Strong governance frameworks are needed to ensure the responsible use of AI in renewable energy projects.’ Read more here.
Read more: Global data centre electricity usage to double by 2030
